PT1772741E - Solid sample nuclear magnetic resonance measuring method - Google Patents

Abstract

In a nuclear magnetic resonance measuring method for solid samples, there are provided the steps of exciting a nuclear spin of a solid sample placed in a static magnetic field by allowing exciting pulses to pass through a receiver coil. A FID signal (Free Induction Decay signal) is received from the excited nuclear spin by applying reading pulses after a lapse of time Ä after applying the exciting pulses is finished. An Inversion Recovery IR - Nuclear Magnetic Resonance NMR spectrum is obtained by subjecting the FID signal to a frequency conversion process. The estimation of a constituent ratio of constituent substances of the solid sample is achieved by acquiring a plurality of FID signals while changing the period of time Ä, by calculating IR-NMR spectra based on these FID signals, by obtaining a longitudinal relaxation curve by plotting spectrum strength with respect to time Ä in a specific frequency of the TR-NMR spectrum, and by making a regression analysis while regarding the longitudinal relaxation curve as a total sum of a plurality of longitudinal relaxation curves differing in longitudinal relaxation time.

Description

1

DESCRIPTION

METHOD OF MEASUREMENT OF NUCLEAR MAGNETIC RESONANCE IN SOLID SAMPLE " TECHNICAL FIELD The present invention relates to a method of measuring nuclear magnetic resonance for solid samples in which a nuclear spin of a solid sample placed in a static magnetic field is excited by the use of a receiver coil, and then an FID signal (Free Induction Decay) emitted from the nuclear spin thus excited, and is then subjected to a frequency conversion process, thereby obtaining an IR (Nuclear Magnetic Resonance) NMR spectrum.

TECHNICAL BACKGROUND

WO 99/42852 discloses a nuclear magnetic resonance spectroscopy technique which is applied to solid samples subjected to a translational motion for the chemical analysis of the samples on a conveyor passing through a detector module. The samples are passed through the detector module at a predetermined rate v, where a spatially uniform constant magnetic field B0 is provided, substantially parallel to the velocity direction v and a radio frequency field pulse B transverse to the field B0. Nuclear magnetic resonance parameters selected from the samples passing through the detector module are then used to determine the chemical composition or changes therein. 2

The parameters include the relaxation time Τι which is measured by acquiring an FID signal in an inversion-recovery sequence.

In BM Fung et al .: " Line Narrowing in Solid-State Proton NMR with Acquisition Delay ", Journal of Magnetic Resonance Series A, Vol. 123, 1996, pages 58-63, a line narrowing is described in NMR spectra of solid state proton acquired in a single-pulse experience with acquisition delay. The wide peaks are filtered by discarding the initial portion of the free induction decay before performing the Fourier transform. The authors have examined the proton NMR line narrowing phenomenon of organic solids by studying several amino acids, their N-deuterated analogs, and some aromatic compounds. The effect of the acquisition delay on the 19F NMR of CaF2 and that on NMR of organic solids are also compared.

In recent years, a near infrared (NIR) spectroscopy method, a powder X-ray diffraction method, or a solid CMR method have been used as methods for physically measuring a solid sample. However, such measurement methods are disadvantageous, for example, because a quantitative analysis can not be performed without references because the limit of detection is high because the signal intensity depends on the size of the crystal and because a specific crystal form, such as an amorphous form, can not be detected.

On the other hand, a proton NMR (PMR) method has been widely used as a means to measure a dissolved sample in a solution. 3

Protons have a high natural abundance ratio, and have higher detection sensitivity than other elements, and are therefore suitable for analysis. The PMR method is performed so that the protons placed in a static magnetic field are irradiated with an RF magnetic field, and wherein the energy variation of the resonant protons with the RF magnetic field is recorded as an electrical signal. The principle of proton nuclear magnetic resonance is as follows.

An atomic nucleus has an instant magnet (spin magnetic moment). The spin magnetic moment of a proton placed in a non-magnetic environment faces a random direction. When it is placed in a static-magnetic field (+ Z direction), the magnetic moment begins the Larmor precession at a slightly oblique angle to the HO axis of the magnetic field. Its angular velocity ωΟ is proportional to the intensity of the magnetic field HO.

ωΟ = (Τ / 2Π) HO The symbol T is called a gyromagnetic ratio, and is an intrinsic constant of a nuclide. The spin phase is in disorder, and is evenly distributed in a vertically conical form.

Groups in the upward direction are excessive in the magnetic field, and the vector M resulting from these groups follows the + Z direction. These are spin groups which are treated as an object of the NMR phenomenon.

To obtain an NMR signal, a radio wave having the same angular velocity as that of the precession motion is radiated from the X axis.

As a result, the spin groups absorb the energy of the radio wave to cause the Mx and My vector components. An NMR signal can be obtained by detecting the My vector component through the use of a receiver coil placed in the y direction.

If the radio wave is a pulse, reference is made as irradiation of, for example, a 90 degree pulse or a 180 degree pulse. Slope angle values, such as 90 degrees or 180 degrees, are specified by an angle at which a spin is inclined from the + Z direction. The tilt angle can be changed by pulse width (microseconds) and pulse strength.

An electric current detected by the receiving coil is called " FID " (Free Induction Decay), and its intensity is maximized when the irradiated pulses are cut, and is attenuated over time. The orbit of a magnetic moment M during a relaxation process is provided by recording the intensity of an electric current produced by a sensing coil when a 90 degree pulse is irradiated. This method of measurement is called IR method (Inversion Recovery Inversion Recovery). 5

Above all, a (180 ° -to-90 °) n-pulse sequence is often used in this method, and is also applied, for example, in a study of the properties of a compound or in magnetic resonance in the medical field. The IR method using this (180 ° -T-90 °) pulse sequence will be explained.

The directions of the magnetic moments of the proton spins in a stationary state coincide with the direction Z. Thus, the resulting vector is present at + Z0. Radiation from a 180 degree pulse allows the direction of the proton spins to make a 180 degree reversal and thus turn in the -ZO direction.

To record this state, a 90 degree pulse is irradiated after a lapse of τ seconds after completing the 180 degree pulse irradiation. The vector takes a position from 180 ° + 90 ° (27 ° degree position) if it is immediately after the irradiation of the 180 degree pulse. In this way, the NMR signal becomes a maximum negative signal.

If the pulse sequence of (180 ° -T-90 °) n is irradiated many times while changing the value of τ, in order to register a change in the NMR signal relative to the value of x, a curve of longitudinal relaxation that results from a single proton and is drawn according to the following equation:

Y = [l-2exp (-t / Tl)] 6 where TI is the time during which the nuclear spin facing the -Z0 direction returns to the initial state of turning in the + Z0 direction, and is called the interlaced spin relaxation time or the longitudinal relaxation time (Tl). FIG. 7 graphically represents this longitudinal relaxation curve. The value of the longitudinal relaxation curve recovers so as to be a zero signal force after 0.693T1 seconds, and reaches a substantially saturated state after 5T1 seconds. The Tl value gives an intrinsic value to a proton environment, and so can be used to obtain information about a molecule. For example, the Tl value reflects a molecule-to-molecule powder distance, and can be used as information that shows a difference in molecular structure.

In a process where the NMR signal is received, an electric current induced by the coil disappears with progression of transverse relaxation immediately upon completion of the 90 degree pulse irradiation. An FID (Free Induction Decay) signal registers this, and is a signal of the spectrum in the time domain when the abscissa axis shows the time. The FID signal is attenuated by exp (- / T2) where T2 is called the transverse relaxation time depending on the environment in which the protons are placed, and is a chemically important piece of information. 7

A so-called NMR spectrum, whose abscissa axis shows a frequency domain, can be detected by subjecting this FID signal to a Fourier transform.

Non-patent literature 1: Journal of American Chemical Society 121, 11554-11557 (1999)

Non-patent literature 2: Australian Journal of Soil Research 38, 665-683 (2000)

Non-patent literature 3: Solid State Nuclear Magnetic Resonance 15, 239-248 (2000)

Non-patent literature 4: Journal of Chemometrics 13, 95-110 (1999)

Disclosure of the Invention

Problems to be Solved by the Invention It is well known that the IR-NMR method mentioned above is not suitable for the measurement of solid samples.

According to the conventional IR-NMR method, a spectrum signal of an FID signal is influenced by the protons of the water molecules contained in a solid sample, so other required proton signals can not appear differently.

Presumably, the reason is that the protons occupying various parts of a crystal molecule have mutually different transverse relaxation times T2 in the solid sample, and therefore signals with a multiplicity of transverse relaxation times T2 are mixed with the FID signal . 8

Thus, even if this is examined according to the IR method, only the FID signal will be obtained in which the transverse relaxation times T2 are mixed together and are a mean of one another. It is therefore an object of the present invention to provide a method of measuring nuclear magnetic resonance for solid samples which is capable of selectively measuring a specific proton in a molecule when an IR-NMR spectrum is measured such that a cell is placed on a receiving coil of an NMR signal and then the solid sample is inserted into this cell. This object is achieved by a method as defined in claim 1.

The above and other advantages, features and effects of the present invention will become apparent from the description given below with reference to the accompanying drawings.

Brief Description of the Drawings [FIG. 1] FIG. 1 is a system block diagram of a metering apparatus that performs a solid sample nuclear magnetic resonance measurement method of the present invention.

[FIG. 2] FIG. 2 is a longitudinal cross-sectional view showing a sample tube 21 in which a sample, which is a solid powder, is placed.

[FIG. 3] FIG. 3 shows an internal structure of an NMR metering chamber 29. 9

[FIG. 4] FIG. 4 (a) shows the waveform of a high frequency signal supplied to a receiver coil 7 and the waveform of a receiver FID signal, and FIG. 4 (b) is a graph showing an FID signal that has been observed and has not yet undergone a phase adjustment.

[FIG. 5] FIG. 5 (a) shows the waveform of the frequency spectrum of a receiving FID signal when the receive delay time Dd is 0, and FIG. 5 (b) shows the frequency spectrum waveform of a receiver FID signal when the reception delay time Dd is 14 μ3θρ.

[FIG. 6] FIG. 6 is a graph of longitudinal relaxation time showing crystal polymorphism analysis of an arginine powder.

[FIG. 7] FIG. 7 is a graph showing the longitudinal relaxation curve resulting from a single proton.

Description of Symbols 2 RF generator 3 Pulse programmer

4 RF Port

5 RF Power Amplifier

6 T / R switch 7 Receiving coil

8 RF-AMP

9 IF-AMP 10 Phase Detector

11 DC-AMP 10 12

Low pass filter

13 A / D Converter

14 CPU 21 Sample Tube 23 Cover 24 Resin Tube 27 Container 28 Condenser

29 NMR measurement chamber 30 Circuit setting plate 31 Terminal

BEST MODE FOR CARRYING OUT THE INVENTION FIG. 1 is a block diagram of an NMR metering apparatus for performing a solid sample nuclear magnetic resonance measurement method of the present invention. The NMR metering apparatus includes an RF generator 2 which generates a series of high frequency signals having a constant frequency (300 MHz, for example), a pulse programmer 3 which generates a pulse signal for modulation, such as a pulse a RF port 4 which applies the pulse modulation to a high frequency signal generated by the RF generator, and an RF power amplifier 5 which amplifies a high frequency signal which has been subjected to modulation up to a few tens of watts.

A high frequency pulse signal amplified by the RF power amplifier 5 passes through the T / R switch 6 in a transmission mode and is applied to the receiver coil 7. The high frequency pulse signal applied to the receiver coil 7 is irradiated in a sample inserted into the receiver coil 7.

Resulting from the high-frequency pulse signal irradiation, an RF electric current produced in the receiver coil 7 by changing the proton spin of the sample passes through the T / R switch 6 in a receiving mode, then passes through the RF- AMP 8 and IF-AMP 9, and enters the phase detector 10. This received signal is thus a time domain FID signal. The FID signal which has been subjected to phase detection by the phase detector 10 becomes a signal present in a range of audible frequencies. After this FID signal passes through the DC-AMP 11, an RF component of the FID signal is removed in the low-pass filter 12. After passing through the A / D converter 13, the FID signal is input to the CPU 14 in the form of a signal time domain, and is stored in a given memory. The time domain digital signal stored in the memory of the CPU 14 is subjected to a Fourier transform in the CPU 14, and then becomes a so-called frequency domain NMR spectrum signal. FIG. 2 shows a sample tube 21 in which is placed a sample which is a solid powder. The sample tube 21 comprises a thick glass tube 22 and a resinous cap 23 with which the tube inlet is closed. For example, tetrafluoroethylene resin may be used as the above-mentioned resin.

To place a sample in the sample tube 21, the sample is first inserted into the glass tube 22, and then the inlet of the glass tube 22 is closed with the cap 23. The cap 23 12 has a small hole through which it passes through Water. A resin tube leading to a dehydrating agent described below is attached to this small hole. FIG. 3 shows the inside of an NMR metering chamber 29. The NMR metering chamber 29 is filled with a gas, such as air or nitrogen. A magnetic field H penetrates vertically into the measuring chamber. The sample tube 21 including a sample is placed obliquely in the NMR measurement chamber 29. A resin tube 24 is inserted into the cap 23 placed in the sample tube 21.

A variation in temperature influences the longitudinal relaxation time, and it is therefore recommended to control the sample tube 21 so that the sample tube 21 maintains a constant temperature. In addition, it is necessary to measure all the samples, among which one will attempt a comparison, at the same temperature.

To measure a water-susceptible sample, a vessel 27 containing a desiccant is placed in the upper part of the NMR metering chamber 29. The end of the resin tube 24 of the sample tube 21 is connected to the vessel 27 containing the desiccant. Likewise, a hydrate can be prevented from being produced during measurement, and data in which the influence of added water is negligible can be obtained. The sample tube 21 is positioned so that the sample coincides with the central portion of the wound coil 7 wound 13 as a solenoid. The terminals of the receiver coil 7 are connected to a capacitor 28 of a circuit adjustment plate 30 connected to the chamber 29 and to another terminal 31, respectively. FIG. 4 (a) shows the waveform of a high frequency signal supplied to the receiver coil 7 and the waveform of a receiver FID signal.

A 180 degree pulse signal is first supplied to the receiving coil 7. As a result, the energy state of a proton present in the sample reaches a so-called high energy state, and the directions of the magnetic moments simultaneously make a 180 degree negative direction.

Thereafter, the transmission operation is stopped for τ seconds. The energy level is reduced by the longitudinal relaxation during this period of τ seconds.

Then a 90 degree pulse is irradiated to observe the phenomenon where the proton returns to a steady state.

After irradiating the 90 degree pulse, the metering system is set to a receive mode, and the waveform of a receiver FID signal generated by the receiver coil 7 is observed.

In the present invention, upon entering the receive mode, the computer begins to store the receiving FID signal after a reception delay time interval Dd. Thus, the signals picked up until the reception delay time Dd 14 has ended after entering the receive mode are not stored. The reception delay time Dd is fixed during the measurement.

If the reception delay time Dd is set to be too short, a situation occurs in which the noise mingles with that in the vicinity of the center of the frequency spectrum, and thus a weak signal is hidden.

Therefore, it is recommended to set the reception delay time Dd to be long. If the reception delay time Dd is set to be too long, the signal strength will be weaker. Therefore, the reception delay time Dd is set to be within a suitable range (e.g., 5 to 20 microseconds, preferably 10 to 15 microseconds). Accordingly, the ratio of the strength of the spectrum (i.e., the signal) to an unnecessary signal (i.e., noise) is set to reach a maximum. The stored receiver FID signal is subjected to a Fourier transform by the computer in order to display an NMR signal waveform in the frequency domain. This NMR signal waveform is produced and recorded. The above-mentioned (180 ° -x-90 °) pulse sequence may be applied over a space of τ once or n times. If a pulse sequence is applied only once, the time domain digital signal introduced into the CPU 14 is subjected to an unchanged Fourier transform. On the other hand, if the pulse sequence is applied n times, the digital time domain signal introduced in the CPU 14 is averaged n times, and is then subjected to a Fourier transform. The latter case is preferred, given that values apart from a mean value can be excluded by averaging the FID receiver signal despite consuming longer measurement times.

Upon completion of the application of the (180 ° -t-90 °) pulse sequence one or n times, another pulse sequence of (180 ° -T-90 °), at which the τ value has been exchanged, is again applied one or n times. The pulse sequence of (180 ° -θ-90 °) is applied in this way while changing the value of τ gradually from zero to a period of about five times the longitudinal relaxation time T1.

A waveform of the receiver FID signal where the time τ is a variable and can be obtained by this process. The computer submits the waveform of the FID receiver signal to a Fourier transform, and obtains a frequency spectrum waveform.

A desirable method of signal processing will be described herein. It is permissible for the FID signal obtained as above to be multiplied by a window function, such as an Exponential or Gaussian, having adequate force in order to remove a noise component, and then a Fourier transform is applied.

In most cases, the frequency spectrum obtained by the Fourier transform is accompanied by a distortion at the baseline. The observed FID signal is as in FIG. 4 (b) in aspect. However, this is obtained by observing the rotational motion of a magnetization vector from one direction. The movement is in fact one of rotation. What is required to convert this to a spectrum is an angle to a first observation point from an observation direction (order phase value 0; PhCO) and an angle from the first observation point to a second observation point of phase of the order PhCl). A correct frequency spectrum can be obtained by subjecting this to a Fourier transform.

Therefore, an attempt of phase adjustment is made in relation to the FID signal using a phase value of the intrinsic order of a measuring condition and a phase value of order 0 which can be defined approximately from the force of the two ends of a spectrum. The phase value of Ia order herein may be obtained by adjusting the phase so that an extreme change in the shape of the spectrum does not occur in a range in which the influence of a return signal is exerted and in a range in which the influence of a signal is not exerted by the use of a spectrum measured with a suitable digital filter. Once this value is set, the same value can be used unless the take-in measurement condition is changed.

Unlike the phase value of Ia order, the phase value of order 0 is not constant, depending on the measurement. There is a possibility that this will be gradually and continuously changed even during a series of measurements. Therefore, a 0-order phase adjustment is performed precisely in relation to a spectrum in which the phase adjustments of the 1st order and 2nd order have been performed by attempts.

This precise order 0 phase adjustment is performed simultaneously with the calculation of a distorted baseline of a spectrum. Let the real part of the spectrum that has undergone a phase adjustment be ro, let its imaginary part be i0, and let the order value 0 be corrected from the current state which is PhCO. The real part of the spectrum that has undergone the true adjustment phase is expressed as follows.

For example, a baseline that makes a sin curve that can be seen as resulting from the imprecision of the signal strength that occurs immediately after the start of a take-in operation can be applied as a distorted baseline. In this case, the baseline basl is expressed with respect to the frequency x as follows.

Basl = A + Bsen (Cx + D) The sin curve was mentioned as an example, and therefore other baseline functions can be used in the same way according to the circumstances.

The five coefficients PhCO, A, B, C, and D are calculated according to the least squares method so that the two values (r and basl) coincide with each other in the best way on the outside of the spectrum. If the sample has difficulty in finding a specific baseline pattern, a correction phase may be performed such that the outside of a typical frequency spectrum 18 is viewed as a baseline pattern and other frequency spectra becomes analogous to this.

Since the calculated PhCO value thus fluctuates through the influence of noise, corrections are made by estimating from the preceding and subsequent values in order to create a continuous variation, and a precise 0-order phase value is defined. A phase correction is performed using this value, and a frequency spectrum corresponding to each time x is obtained.

If the baseline corrections were performed using the value of a calculated variable for the baseline, a frequency spectrum can be obtained which can be seen as " standard powder " gauss standard ", excluding a case in which there is a specific dipole-dipole interaction.

However, since the term baseline uses many variables, errors are easily produced as a result of the influence of noise. In addition, since the amount of baseline compensation is proportional to the total signal strength, no change will be noticed in an analytical result of the longitudinal relaxation curve even if baseline corrections are not made.

Therefore, if the error becomes large, it is allowed to proceed to the next step of the process without performing baseline correction. If a sample has difficulty finding a specific baseline standard, baseline correction can not be performed, but, as above, there is no need to perform baseline correction. There is one case where a frequency spectrum in which the phase or the like has been corrected still includes a non-negligible noise. In this case, a softening digital filter suitable for the frequency spectrum is used in order to remove the noise. An FFT filter, a Savitzky-Golay filter, or a moving average filter may be referred to as softening digital filters which may be used herein. If the FFT filter is used here, this is equivalent to having a smoothing process applied to a time domain spectrum (FID Signal). The smoothing process can be performed when the time base spectrum is obtained.

Examples of the frequency spectrum waveform of a received FID Receiver signal are shown in FIG. 5 (a) and FIG. 5 (b). FIG. 5 (a) and FIG. 5 (b) are graphs showing the comparison of the frequency spectrum waveform that appears when the receive delay time Dd is 0 and the frequency spectrum waveform that appears when the receive delay time Dd is defined to 14 microseconds in the frequency spectrum waveform of a FID IR-NMR receiver FID signal. The ordinate axis shows the strength of the spectrum, while the abscissa axis shows the relative frequency (unit: ppm) based on the peak of the signal (4.5 ppm) of a water proton.

Four waveforms of FIG. 5 (a) are waveforms of the frequency spectrum that appear when the receive delay time Dd is 0.

Four waveforms of FIG. 5 (b) are waveforms of the frequency spectrum that appear when the receive delay time Dd is 14 microseconds. 20

In FIG. 5 (a) and FIG. 5 (b), the most left cL waveforms appear when the PW pulse amplitude of a 90 degree pulse signal is 2 microseconds, the second left waveforms appear when the PW pulse amplitude of a pulse signal of 90 degrees is 5 microseconds, the third left waveforms appear when the PW pulse amplitude of a 90 degree pulse signal is 10 microseconds, and the rightmost waveforms appear when the PW pulse amplitude of a 90 degree pulse signal is 13 microseconds.

If the pulse amplitude PW is 13 microseconds functions more properly as a pulse signal whereby a proton spin is rotated 90 degrees, the description will hereinafter be given as attention to the rightmost waveforms of FIGS. 5 (a) and 5 (b).

A multiplicity of waveforms are drawn as waveforms to the right of FIGS. 5 (a) and 5 (b). These are different from each other at time τ from the end of the application of a 180 degree pulse signal to the start of the application of a 90 degree pulse signal. In this graph, time τ has some dozens of time phases ranging from substantially zero seconds to 70 seconds.

In FIG. 5 (a), at any time instant τ, spectrum waveforms, other than protons, such as water, do not appear prominently. In other words, many of these are proton signals of water near 0 ppm. However, in FIG. 5 (b), the waveforms of the spectrum having a peak at frequencies other than that of the water proton 21 appear on each side of the NMR spectrum of the water proton.

With respect to these spectra having left or right side peaks, respectively, of the water proton NMR spectrum, the time τ is defined on the abscissa axis, and the force of the spectrum is placed on the axis of the ordinates, thus creating a longitudinal relaxation curve. The longitudinal relaxation curve of the selected proton is often called a " longitudinal relaxation curve selected " The selected longitudinal relaxation curve is a curve having an inclination which differs from the longitudinal relaxation curve of the water proton, and thus is distinguished from the longitudinal relaxation curve of the water proton.

A multiplicity of proton types contained in a solid sample can be quantified by recording the selected longitudinal relaxation curve and doing a regression analysis.

This method, which is not part of the present invention, is carried out assuming that (1) the measured individual sample is a mixture comprising a multiplicity of component substances differing in crystal form, and (2) the proton of each substance component has different TI longitudinal relaxation time.

Assuming that the selected longitudinal relaxation curve is formed by the linear sum of a multiplicity of longitudinal relaxation curves differing in the longitudinal relaxation time T1, the adjustment is performed using a regression curve, thus making it possible to automatically obtain the value and proportion structural curve of the longitudinal relaxation curves that form the selected longitudinal relaxation curve. The regression analysis technique, which is not part of the present invention, will be described in more detail below.

At the specific frequency of a frequency spectrum obtained, the signal strength corresponding to variable dwell time (τ) is plotted so that the selected longitudinal relaxation curve mentioned above is obtained.

Preferably, it is used as the specific frequency, not a frequency close to about 0 ppm, as described above, but a frequency whose force is maintained to some extent and has little influence on the water signal or other noises. &Quot;

Additionally, it is permissible to use an integral value of the signal strength that falls within a specific frequency range, rather than the signal strength at a specific frequency. In this case, in the same way, a smoothing process can be seen as having been carried out herein. The abundance rate of the constituents is determined by doing a regression analysis on the assumption that the selected longitudinal relaxation curve obtained herein is the sum of the constituents having their respective Tl. The same mathematical formula used is as follows: [Formula 1]

where Tl is the longitudinal relaxation time, -Ho is the initial force immediately after the application of a 180 degree pulse, f is a coefficient (ie, a signal strength ratio that is equal to the abundance ratio of the constituents) , i is a suffix showing each constituent, en is the total number of constituents.

Ideally, H0 here is 2. However in practice H0 is slightly smaller than 2 due, for example, to the imprecision of a 180 degree pulse. Although a crucial problem does not occur even though the analysis is done in this state, it takes a lot of time for the convergence of the least squares method. Therefore, the equation f '= fHo / 2 is formulated, and then transformed, thus obtaining the following formula. {Formula 2) M * ti

Here the second term is a constant term not dependent on time x, and thus a calculation can be performed by fitting this term to a constant C. 24

Additionally, provided it is suitable for the x-axis showing the time τ use a logarithmic representation as shown in FIG. 6, the equation s = lnx is formulated and transformed. As a result, the following formula is obtained. Formula 3]

1A-2e -> + C such that Si is a logarithmic value of the longitudinal relaxation time Tli.

Si = lnTli

Using this formula, optimization is performed according to the least squares method, and the longitudinal relaxation time of each constituent and its coefficient are calculated. The proper coefficient f and the appropriate H0 value can be calculated from f 'and C. In most practical cases, the value of H0 has little difference in each constituent, and so the substitution can be performed between f' and f .

If the constituents are significantly different from each other in the longitudinal relaxation time T1, and are sufficient in their respective amounts, the Tl values and the abundance ratio of each constituent can be obtained with certain degrees of accuracy even though the calculation is only one sample.

However, if the constituents were similar to each other in the Tl longitudinal relaxation time, or if the specific constituents of a sample to be measured are in extremely small amounts, the error will become large. To avoid this, an analysis is performed according to the non-linear least squares method for a multiplicity of samples containing the common constituents. This makes it possible to increase the accuracy of the IT value of each constituent and measure its abundance ratio with high precision.

Examples (not part of the present invention) < Example 1 >

According to the IR NMR method of the present invention, the quantitative capacity of an amorphous substance mixed with an arginine powder was evaluated.

This was measured with an INOVA 300 NMR meter made by Varian, Inc. Arginine was used as the sample. FIG. 6 is the longitudinal relaxation time plot showing a crystal-polymorphic analysis of the arginine powder. The axis of the ordinates shows the signal strength, and the abscissa axis shows the time τ (log) varying from the end of the application of a 180 degree pulse to the start of the 90 degree pulse application. The reference character a represents a pulverized product obtained by spraying a crude anhydrous arginine powder using an agate mortar, the reference character b represents a sample in which the pulverized product is added to a 70% crude arginine powder, the reference character c represents a sample in which the pulverized product is added to the crude arginine powder by 50%, the reference character d represents a sample in which the powdered product is added to the crude arginine powder by 20%, and the reference character and represents the crude arginine powder.

The graphs are based on the sum of the FID signals of the constituents that differ from the relaxation time.

Three types of powders, i.e., arginine crystals, amorphous arginine substances, and aggregates are contained in these powder samples. Let the longitudinal relaxation time TI of the arginine crystal be represented as Tla, let the longitudinal relaxation time TI of the amorphous arginine substance be represented as Tlb, and allow the relaxation time TI of the aggregate to be represented by Tlc. The strength of the Gtotal signal can be expressed as follows.

< 3101 to 1 g + 1 GB + 1 gC-1 aU-2exp (-τ / Tla) + fb {1 ~ 2exp ('/ tb) }}

In this equation, fa, fb, and fc represent the ratios between the constituents.

If the measurements of a multiplicity of τ are performed, the various equations which are mentioned above can be obtained. Since the simultaneous equations in which Tla, Tlb, Tlc, fa, fb, and fc are unknown, can be formulated, Tla, Tlb, Tlc, fa, fb, and fc can be determined by solving the equations. A statistical method, such as the non-linear least squares method or the maximum likelihood estimation method, may be used in proportion to an increase in the number of measurement points, and thus obviously the accuracy is improved.

Based on the graphs of FIG. 6, the simultaneous equations were solved, using the statistical analysis. As a result, the following values were obtained.

Tla = 28.02 seconds Tlb = 12.01 seconds Tlc = 3.99 seconds

In the powdered product obtained by spraying crude anhydrous arginine powder using an agate mortar, the ratio of the constituents was as follows.

Fa = 17.80%,

Fb = 64.46%,

Fc = 17.73%

In the sample where the pulverized product is added to a 70% crude arginine powder, the ratio of the constituents was as follows.

Fa = 41.01%

Fb = 46.19%,

Fc = 12.80%

In the sample where the pulverized product is added to the crude arginine powder by 50%, the ratio of the constituents was as follows.

Fa = 57.52%,

Fb = 32.42%,

Fc = 10.06%

In the sample where the pulverized product is added to the crude arginine powder by 20%, the ratio of the constituents was as follows.

Fa = 79.94%,

Fb = 15.82%,

Fc = 10.06% 28

In the crude arginine powder, the ratio of the constituents was as follows.

Fa = 97.10%,

Fb = 2.55%,

Fc = 0.35%

As described above, the step of the decomposition process in an amount of the curves and the step of the adjusting process are applied to a number of types of mixtures which are the same in the composition of the chemical compound but are different in crystal form. As a result, the longitudinal relaxation time and the ratio of selected proton constituents can be obtained, respectively, in which a signal deriving from each crystal form is given. < Example 2 > (1) Preparation of samples for measurement

Indomethacin was used as the sample. A biochemical reagent from Wako Pure Chemical Industries, Ltd. was purchased as indomethacin. Based on this, the samples were prepared according to the seven methods mentioned below. - Sample 1 (Recrystallization from MeCN; MeCN) When heated, 2 grams of indomethacin was dissolved in MeCN (acetonitrile) (50mL). The undissolved crystals were filtered and removed, and the filtrate was allowed to stand at room temperature. After one day, the precipitated crystals were filtered, and a wash with MeCN was performed. Then, it was dried under reduced pressure. Sample 2 (recrystallization from Et 2 O, Et 2 O)

Upon being warmed slightly, 1 gram of indomethacin was dissolved in Et2O (diethyl ether) (50mL). The undissolved crystals were filtered and removed, and the filtrate was allowed to stand at room temperature. After three days, the precipitated crystals were filtered, and a wash with Et 2 O was performed. Then, it was dried under reduced pressure. - Sample 3 (EtOH-water, no recrystallization time; Oh)

Upon being heated slightly, 2 grams of indomethacin was dissolved in EtOH (ethanol) (50mL). The water was gradually added to this solution while stirring. The water was no longer added when the crystals began to be precipitated. Immediately, a large amount of crystals were precipitated, and the system reached a state where it could no longer be agitated. Immediately, the precipitated crystals were filtered, and washed with 50% EtOH. Then, it was dried under reduced pressure.

Sample 4 (EtOH-water, 18 hours recrystallization time, 18h)

In the same manner as for Sample 3, the crystals were precipitated. The solution being in a non-stirring state was continuously agitated with a magnetic stirrer without being stopped. Although this in the beginning was in a state of not being able to be agitated it was gradually shifting to a state of being able to be agitated. This was stirred for 18 hours at room temperature. The crystals were then filtered, and washed with 50% EtOH. Then, it was dried under reduced pressure. Sample 5 (OM)

Part of the crystals obtained in Sample 3 were sprayed with an agate mortar, and a powdered product was obtained. Sample 6 (18M)

Part of the crystals obtained in sample 4 were sprayed with an agate mortar, and a powdered product was obtained. Sample 7 (R) The reagent purchased was used in an unchanged form. (2) Measurements

These seven samples were placed in a 5mmO NMR tube so that the height was about 25 to 35mm, and then placed in a vacuum desiccator with diphosphoric Pentaxide (P205) used as the dehydrating agent for one or more hours, and drying was performed. These were removed from the desiccant immediately prior to measurement, and were immediately sealed. These were used as samples for measurement. Measurement was performed using a Bruker DPX-300 / 5mm double probe < t > CH / SADC + A / D converter / XWIN-NMR software. The correction was adjusted using another NMR tube in which CDC13 was placed to the same sample height, and then the target sample was placed on the probe. The measurement was performed in a state of " SWEEP OFF, " " SPIN OFF, " and " LOCK OFF. " The temperature of the sample was adjusted to 23 ° C using a temperature controller. The tlir 31 pulse program supplied as standard by the Bruker Company was used, and the measurements were performed with the following parameter values. As a result, a time base spectrum (FID signal) was obtained.

Dl (relaxation delay): 90sec Pl = pw (pulse 90o): 8.45 με P2 = pl (180o pulse): 16.9 με DE = Dd (reception delay time): 16 με DEI of delay ranging from the closing of the transmission coil to the transmission coil opening): 3 με 01 (observation center frequency): 2,54 ppm NS (operation integration number): 8 DS (scan model): 2

SW (observation amplitude): 497,314ppm DigMod (digitizer mode): Analog ParMod (parameter mode): 2D SI (data size): [F2] 16384, [Fl] 64 (F1 and F2 are observation axes in an NMR bi-dimensional) TD (size of acquired data): [F2] 16384, [Fl] 45

A Fourier transform (xf2) was applied to the time base spectrum obtained as above with the following parameter values. As a result, a frequency spectrum was obtained for each x time period.

WDW (Window Function): EM LB (Line Magnification Factor): 300Hz

PhCO: -99.56 (value at which the forces at both ends of a spectrum are substantially the same: value of the variable depending on each measure)

PhCl: 130 (values analogous to a spectrum measured in digital digitizer mode: constant value) 32

Since this spectrum is a two-dimensional NMR spectrum interrelated with spectra corresponding to periods of dwell time (τ), " split2D " was performed, and the real part and the imaginary part of one-dimensional spectra corresponding to the periods of waiting time (τ). The data relating to these one-dimensional spectra were copied from a measuring computer to a computer data processor. (3) Data processing The real part (r0) and the imaginary part (io) of the one-dimensional spectra copied to the data-processing computer are shown by arranging the Y-axis values from the low-wave-side side to the of the high wave number. Although the values on the X-axis (frequency axis) range from 251.17ppm to -246,124ppm, there is no need to make a calculation using especially the ppm unit. Therefore, the integers x (ranging from 1 to 16384) have been replaced. The real part (r) of a frequency spectrum adjusted by a correct 0-order phase value is expressed as follows: R = r0cos (PhCO) -i0sen (PhCO) where PhCO is a phase value of order 0 to be corrected from the current state.

On the other hand, the baseline of a sin curve resulting from the inaccuracy of the signal strength appearing immediately after the onset of admission was used as the baseline. In this case, with respect to x, the baseline (basl) is expressed as follows.

The coefficients PhCO, A, B, C, and D were calculated according to the non-linear method of the least squares so that orb basl coincide with each other in as the X-axis values range from 1 to 2048 and insofar as the X-axis values range from 14336 to 16384. In the condition of r = basl, the following equation can be formulated. (PhCO) - i0sen (PhCO) = A + Bsen (Cx + D)

However, since calculations can not be performed in this state, the equation was transformed as follows: r0 = {r0sen (PhCO) + A + Bsen (Cx + D)} / cos (PhCO) where ro is a variable dependent, and io ex are independent variables. The calculation of the results of sample 1 (recrystallization of MeCN) is shown below. 34 [Table ΙΑ] ABCD PhCO PhCO (corrected) -18560088,34 -35371960,82 0.530416 8164,089 -2.53131 -2.6 -8641245.5 -35084197.73 0.548123 8164.511 -2, 45099 -2,475 -18880570,91 -34750676,41 0.529324 8136,935 -1.91209 -2.35 19188701.9 -34621817.61 0.520866 8161.021 -2.54596 -2.225 -18995107.28 -34549265 , 62 0.514546 8160.592 -2.71764 -2.1 -19006363,07 -34423089,96 0.523416 8150.432 -1.99075 -1.975 -19172530.65 -33916071.82 0.526752 8141.717 -2.09543 -1.85 -18973751,48 -34014594.16 0.527977 8148,807 -1.77497 -1.725 -18813541.07 -33853800.51 0.518876 8148.219 -1.28569 -1.6 -18475271,09 -33889180,77 0.526671 8156,595 -1.46661 -1.475 -18661080,21 -33385697.97 0.519225 8132.685 -1.09835 -1.38 -18329245,09 -33258451,88 0.519425 8152.947 -0.922477 -1.285 -18171542,97 -32592696.84 0.527473 8165.525 -1.36935 -1.19 -17898423.23 -32232619.94 0.500092 8178.881 -1 , 81755 -1.095 17778341.99 -31700091,17 0.524117 8148.15 -0.657729 -1 -17362809.5 -31264583.21 0.529787 8167.705 -1.29616 -0, 905 -17073689,39 30468116,46 0.533883 8174.173 -1.36431 -0.81 -16883124,64 -29690471,43 0.527328 8164,502 -0.335533 -0.715 -16322519,58 -29085342,8 0.50976 8176.524 -0.852706 -0.62 -15921714,35 -28259108.35 0.517755 8172.68 -0.20242 -0.525 -15039798.62 -27016303.73 0.525881 8177.689 -0 , 537248 -0.43 35 [Table 1B] ABCD PhCO PhCO (corrected) -14038237,56 -25333734,36 0.51251 8196.356 -0.774681 -0.335 -12874742.1 -23324000.45 0.502681 8174, 393 -0.201961 -0.24 -11578879,81 -20837380 0.483257 8193.796 0.0757171 -0.145 -10312676.49 -18450827.93 0.504593 8209.986 -0.477082 -0.05 -8961944 , 001 -16532052.95 0.493431 8183.218 0.534294 0.045 -7342144.464 -13461109.88 0.509912 8233.159 -1.36166 0.14 -5848768,123 -10527056.88 0.506645 8195, 595 0.731044 0.235 -3842273,808 -7177118,849 0.484079 8267.187 -1.67545 0.33 -2224252.287 -3774084.28 0.48088 8265.813 0.736975 0.425 276076.62 -469574, 79 0.496588 8540.988 5.96791 0.52 2536056.32 4267782.632 0.527585 8190.665 1.10551 0.59 5100 157,192 8696889,756 0.528072 8265,608 -1.41766 0.66 7626494,571 13138814,58 0.501366 8140.71 3.00361 0.73 10207851,23 17406297.9 0.520495 8187,727 1.37506 0.8 12755338.8 22248723,49 0.513167 8213.313 -0.560938 0.87 15217272.14 26496594.42 0.501938 8170.271 1.46622 0.94 17509796.55 30915702.06 0.524265 8165 , 398 0.978927 1.01 19843967.81 34968657.81 0.516925 8139.288 1.17099 1.08 21569809.08 38350446.16 0.528953 8146,715 0.920894 1.15 23629600.85 41782617,37 0.514997 8126.39 1.53121 1.22 25851219.25 46057622.54 0.534418 8128.976 1.29237 1.29 26746444.55 48138237.01 0.524036 8102.483 1.28472 1.36 27052469, 48 48726429.24 0.548169 8123.783 1.21352 1.43 27122712.85 48877451.18 0.546263 8120.23 1.49238 1.5

The tables show that the PhCO value is gradually and continuously changed according to the order of measurement but fluctuates under the influence of noise. Especially near the point where the signal strength becomes weak, there occurs a wide variation. The (corrected) PhCO value was produced being corrected from all PhCO changes so that the PhCO value can be changed continuously. An adjusted order 0 phase frequency spectrum was obtained using the PhCO (corrected) values. The 191-point FFT smoothing filter was applied to the frequency spectrum obtained in order to remove the noise. The signal strength at which the X-axis value of the frequency spectrum is 7000 (corresponding to 38.72ppm in terms of frequency) was plotted against each time-out period (τ), thus yielding curves of longitudinal relaxation.

Likewise, the longitudinal relaxation curves for the indomethacin samples from sample 2 to sample 7 were produced. These results are shown in Table 2A and Table 2B.

Table 2A ze τ Ιη, τ MeCN Et 20 Oh 18h 0M 18M R 0.00005 -9.903 -291161455,2 -155516618,4 -195626262 -197872856 -253287019 -260896939,1 -227236771,5 0,0001 -9,210 -287450816,6 -154641382,2 -194018101,7 -195819871,1 -249922140,7 -258510835,6 -226522843,2 0,0002 -8,517 -281344377,5 -152430772,4 -190327626,7 -191871823,6 -245044259,8 - 253527089,5 -223046648,4 0.0005 -7,601 -274429701 -149344415,7 -186522810,9 -187465980,7 -238631963,6 -247233776,9 -217825802,4 0.001 co -269713709,6 -147473701 -182854460,9 -184959231,7 -234809477,6 -242991257,7 -213448261,4 0.002 -6,215 -266507526,4 -146414134,7 -181081775,2 -182927323,3 -231304494 -241168447,4 -211060199,8 0.005 -5.298 -263771028 , 2 -145861644.9 -179468919,1 -181353321,9 oo co C1-4 0 1 -1 -238154167,7 -209590173,4 0.01 -4,605 -261952033,9 -145839820,3 -178328612,1 -180122264 , 9 -227038461.3 -236566551.9 -208769297.1 0.02 -3.912 -261004660 -145054946,7 -176664136,1 -178937828,9 -225065022,9 -235220506.1 -207285451 0.033 -3.411 -259 380936.4 -144372033,5 -174632980,2 -177309080,7 -222589287,8 -232909951,8 -205286416,1 0,05 -2,996 -258089742,7 -143027430,8 -173070686 -175611920,7 -220082327,4 -229919496,4 -203493156 0.07 -2,659 -255915904.8 -141302210 1 o O Oo -173160692,7 -216547292,6 -227243395,8 -201030182,6 0.1 -2,303 -252558368,2 -139075136,2 -165678870,2 -169694150,5 -211476905,8 -222864377,9 -197455248,7 0,13 -2,040 -249351362,2 -136821459,8 -162266995,6 -166476118 -206458602,5 -218212747 -193623291,2 0 , 16 -1.833 -245690207.6 -134617841,7 -158071337,4 -163140396 -201532962,8 -213567422,9 -190114080,3 0.2 -1,609 -241739614 -131850496,7 -153164145,9 -159100213,6 - 195179877,9 -207739278,5 -185151463,6 0.24 -1.427 -237071074,6 -129002182,6 -147762183,9 -154955205,7 -188755545 -202053699,8 -180517429,7 0,28 -1,273 -23,249,160, 7 -126590928,2 -142693647,7 -151347236,7 -182312056,5 -196685632.9 -176390969,9 0,34 -1,079 -22,6530,741.8 -122506058,9 -135328632,8 -145093739,8 -173840539,5 -168217185.8 1 00 1 Γ- , 1.04 co. 1 ^ 0.4 -0.916 -220496598,5 -118318213,8 -128261043,3 -139201347,2 -164900824,2 -179953499 -162400340,1 0,5 -0,693 -210038795,6 -112314353,6 -117415389,9 -129808047,4 -150893581,5 -167167389,3 -151662970,4 0,63 -0,462 -197499361,2 -103905668,9 -102809053,9 -118038844 -133448745,4 -150662170,8 -137809570,9 0,8 -0,223 -181076704,7 -93574347,3 -85997467,28 -103701448,4 -112192770,8 -130714744,4 -120549226,8

Table 2B 38 τ Ιητ MeCN Et 20 Oh 18h 0M 18M R 1 0.000 -163052730,3 -82572617,16 -68018349,08 -87338421,98 -90101799,73 -107868453,9 -101871818,7 1,2 0,182 -146219540,4 -72259015,57 -52296773,25 -72415023,95 -68645559,61 -86731475,69 -84142928,82 Μ 0,336 -130429744,4 -62224626,16 -37338173,13 -57900768,95 -49448173,08 -67270102,91 -67683849.77 U 0.531 -107216940.4 -48286871,13 -15907679,53 -38193115,42 -23506193.25 -39513814,99 -44201153,32 2 0.693 -85652004,03 -35606036,24 3063811,899 -19974413, 04 360225,831 -14994519.5 -23443790.2 2.4 0.875 -59429812.6 -20289857,14 24722464.63 2028792.872 28223968.75 15309573.61 2534840.572 2.8 1.030 -35810852,15 -6298542, 604 44790809.66 21815205.16 53141500.18 42211017,64 25703379,76 3.33 1,203 -6425363,1 10429300,48 67838343,59 45665053,29 82470254,53 73846925,24 53192628,68 4 1,386 26608757,1 29171169,29 93103745,91 71460688,13 113614659 108586247,6 83381628,65 4.8 1,569 61599886,96 48709256,46 118750824,4 97456406,38 145168962 143681 567.8 113753718 5.2 1.740 96060734 67663962.02 142544133.7 122015774.4 175138741.8 175977632.8 142612124.9 8.7 1.902 129556036.9 85920534.29 163673539.9 144815668.4 202542066.2 205149165.5 169928428 , 8 8 2,079 166279062.8 105272141.3 186941178.3 167881944.9 230768402.2 235242754.1 197704459.5 9.5 2,251 200840690.1 123732949.8 205921088.9 189114146.4 255301839.6 261665671.7 222220472,9 11.3 2.425 233806304.4 140973859 223687312.2 207331682.5 276458523,2 284786709.6 244120946.2 13.5 2,603 265810668.6 156745431 237976913.3 223014838.6 295293492.7 304287268.8 263313224.2 16 2,773 292649624, 3 169730109.8 249699436.2 235171470.5 30864988.2 317704746.8 27742L248.3 20 2,996 320948006 183470639.5 258523553.9 246804516.7 320189820.1 331193025.1 290666621.5 30 3.401 251678877.5 198647558 265509279.8 257194323 , 1 329499842.3 342453754 302536283.5 45 3,807 369516152,1 204765930,9 267931044 259766571,7 332316900,1 345165140,7 306124389.6 60 4.094 373706502.4 206096032.1 268745964 260430164, 7 332548242.5 346389625.2 307157904.5 120 4,787 375306116.8 206846677.7 269562105.2 261225891.1 332984409.5 346825670,7 308183647.7 39

The longitudinal relaxation curves obtained as in Table 2A and Table 2B for seven samples were analyzed simultaneously using the following formula.

[Formula 4] [VÎ ± -2Î ± - H *

Five types of indomethacin constituents were found (in the formula, n = 5). The number n of constituents is determined as follows. First, the value of the 95% security interval of f 'is calculated. If f 'is equal to the value of the safety interval of 95% or less than this value, the constituents are considered to be absent. For example, on the assumption that n = 3, a comparison is made between f'l and the value of the 95% safety interval. If f'l is greater than the value of the 95% security interval, the number n is defined as 4 (ie, n = 4), and an observation is made of when f'l and f'2 are all not less than the value of the 95% safety interval. The number n is increased in this way. If f 'that exceeds an arbitrary number n is less than the value of the 95% security interval, this number n is determined as the number of constituents. Table 3 shows the results of the calculation of the estimated value and the standard error of the longitudinal relaxation time Si.40 [Table 3] [Table 3]

Estimated value Standard error Sl 2.31861 0.01127 S2 1.54191 0.00919 S3 0.07003 0.02001 S4 -6.45511 0.15242 S5 -8.30651 0.10355 Table 4 shows the estimated coefficient results of the calculation fi and the estimated value of the constant value C.

[Table 4] [Table 4] (Estimated value)

MeCN Et20 Oh 0M 18h 18M R fl 179780327,2 76896003,52 26251437,18 35878665,9 30985235,14 27949499,18 43293165,94 f2 123090554,6 86185892,41 163043268,7 165659932,2 206245253,2 234562139,6 199629300, 8 f3 16415415,52 13708383,41 35145641.6 19550687.99 43804157,19 29815831,74 21061569,02 f4 4957365,801 770598,7548 3530400,271 3167832,659 5473638,747 4162542,557 3612612,683 f5 10512716,65 4676184 , 602 5778023,155 6331640,672 B417305,268 8707999,039 6667933,353 C 39738295,64 24610946,24 35708695,18 30340056.96 38601504.15 40693972.15 38463312,15 Table 5 shows the results of the standard error calculation of the coefficient fi and the standard error of the constant C.

[Table 5] [Table 5] (Standard error)

MeCN Et20 Oh 0M 18h 18M R fl 3396240,259 1766837,846 1616393,409 1853087,805 1951895,348 2275001,116 2161095,975 f2 3617216,617 1824354,276 1381613,701 1546611,552 1584035,399 1696369,317 1767513, 852 f3 610013,7184 499394,1109 92968,1109 709505,9899 1125612,37 987598,7324 780590,9893 f4 793692,4788 405463,8619 568153,8707 561671,1539 765325,3398 687375,1105 597531,356 f5 780119,4702 519730 , 4369 650040,1152 630634,2432 807067.8342 708909.6659 660851.0169 C 367194.3623 315202.8009 312050.4795 317748.4315 329143,037 338506,3065 319756.7396 From these results, the values of the (attributed to the type γ), 4.67s (attributed to type a), 1.07s (attributed to amorphous), 1. 57ms (assignment unknown), and 0, 25ms (assignment unknown).

In addition, the constituent ratio of each sample was determined as in Table 6, using the coefficient of force fi. Additionally, the confidence interval of ± 95% was calculated at the same time.

[Table 6] [TABLE 6] TI MeCN Et 20 Oh 0M 18h 18M R 10.16s 53.70 + 2.00% 42.20 + 1.91% 11.23 + 1.36% 10.51 + 1.30% 15, 56 ± 1.58 ± 9.16 ± 1.47 ± 16.08 ± 1.58 ± 4.67s 36.77 ± 2.13% 47.29 ± 1.97 ± 69.75 ± 1.16 ± 69.93 ± 1, 06¾ 71.84 + 1.32¾ 76.86 + 1.10¾ 72.28 + 1.295 l, 07s 4.90 + 0.36¾ 7.52 + 0.54¾ 15.04 + 0.78¾ 14.85 + 0, 75¾ 8.48 + 0.61¾ 9.77 + 0.69¾ 7.62 + 0.57¾ 1.57ms 1.48 + 0.47% 0.42 + 0.44¾ 1.51 + 0.48¾ 1.86 + 0.51¾ 1.31 + 0.48¾ 1.36 + 0.44¾ 1.34 + 0.44¾ 0.25ms 3.14 + 0.461 2.57 + 0.56¾ 2.47 + 0.55¾ 2.85 + 0.54 ± 2.75 ± 0.54 ± 2.85 ± 0.46 ± 2.48 ± 0.48 ± (195 es confidence interval value) 43 44

By this route, the ratio of polymorphic solid crystal constituents of indomethacin can be determined with high accuracy. The longitudinal relaxation time of each constituent, as well as the ratio of the constituent, can be measured at the same time. < Example 3 > (1) Preparation of samples for measurement

Glycine was used as a sample. Special Grade Chemical Reagents from Wako Pure Chemical Industries, Ltd. were used as glycine. Based on these, the samples were prepared according to the eight methods mentioned below. Sample 1 (water recrystallization A)

Upon being slightly heated, 10 grams of glycine was dissolved in water (40mL). The solution was allowed to stand at room temperature. After three days, the precipitated crystals were filtered, and the wash was washed with water. Then, it was dried under reduced pressure. - Sample 2 (water-acetic acid recrystallization; C)

Upon being slightly heated, 15 grams of glycine was dissolved in water (30mL) - acetic acid (3mL). The solution was gradually cooled to room temperature, and the precipitated crystals were filtered. Washing was carried out with water, and the drying was proceeded under reduced pressure. - Sample 3 (water-EtOH recrystallization; B)

Upon being slightly heated, 10 grams of glycine was dissolved in water (40mL). EtOH (20mL) was gradually added while the solution was stirred at room temperature. The precipitated crystals were filtered. Thereafter, the wash was washed with water, and the drying was proceeded under reduced pressure. Sample 4 (Recrystallization water-EtOH drying at 60 degrees B60)

The crystals obtained as for sample 3 were subjected to hot air drying at 60 ° C for 16 hours. Sample 5 (S)

The crystals which had adhered to the wall of the device when the recrystallization operation of the sample 1 was performed were removed. Thereafter, the wash was washed with water, and the drying was proceeded under reduced pressure. Sample 6 (CM)

Part of the crystals obtained in Sample 2 was sprayed with the agate mortar, and a powdered product was obtained. Sample 7 (RM) The purchased reagent was sprayed with an agate mortar, and a powdered product was obtained. Sample 8 (R) The reagent purchased was used unchanged. (2) Measuring

These eight samples were placed in a 5mm NMR > so that the height was about 25 to 35 mm, and then placed in a vacuum desiccator with pentaoxethophosphorous (P205) used as a dehydrating agent for one or two hours, and the drying was carried out. These were removed from the desiccant immediately prior to measurements, and were immediately sealed. They were used as samples for measurement. Measurement was performed using a Bruker DPX-300 / 5mm double probe < g > CH / SADC + A / D converter / XWIN-NMR software. The correction was adjusted using another NMR tube in which CDC13 was placed to the same sample height, and then a target sample was placed on the probe. Measurement was performed in " SWEEP OFF, " " SPIN OFF, " and " LOCK OFF. " The temperature of the sample was adjusted to 23 ° C with the use of a temperature controller. The tlir pulse program supplied as a standard was used by the Bruker Company, the measurement was performed with the following parameter values. As a result, a time base spectrum (FID signal) was obtained.

Dl (Relaxation Delay): 60s

Pl = pw (90 ° pulse): 8.45 s P2 = pl (180o pulse): 16.9 s DE = Dd (reception delay time): 16 s DEI (delay time ranging from the closing of the (number of integration operations): 8 DS (scan model): 2 SW (observation amplitude): 497,314ppm DigMod (scanning range up to the receiver coil opening): 3με 01 (observation center frequency): 2,54ppm Analog ParMod: 2D SI (data size): [F2] 16384, [Fl] 64 47 TD (sample size acquired): [F2] 16384, [F1] and F2 are observation axes in the two-dimensional NMR)

A Fourier transform (xf2) was applied to the time base spectrum obtained as above with the following parameter values. As a result, a frequency spectrum was obtained.

WDW (window function): LB EM (Line Magnification Factor): 300Hz PhCO: -0.64 (value at which the forces at both ends of a spectrum are substantially the same: variable value depending on each measurement)

PhCl: 130 (value analogous to a spectrum measured in digital digitizer mode: constant value)

Since this spectrum is a two-dimensional NMR spectrum consisting of spectra corresponding to time-out periods (τ), " split2D " was performed, and the real part and the imaginary part of the one-dimensional spectra corresponding to the periods of the waiting time (τ) were obtained. Data relating to these one-dimensional spectra were copied from a computer measurement to a computer data processor. (3) Data processing

A 0-order phase adjustment was performed using the real part (ro) and the imaginary part (i0) of the one-dimensional spectrum copied to the data processing computer. The real part (r) of a frequency spectrum adjusted through a correct 0-order phase value is expressed as follows: r = r0cos (PhCO) -i0sen (PhCO) where PhCO is an order phase value 0 to correct from the current state.

On the other hand, the frequency spectrum (b) of the glycine measured separately was used as the baseline.

The coefficients PhCO, A, and B were calculated according to the non-linear least squares method so that oreob coincide with each other as the X-axis values range from 1 to 2048 and to the extent values range from 14336 to 16384.

Table 7A and Table 7B show the results of the calculation of water-acetic acid recrystallization of glycine from sample 2. 49 [Table 7A] [Table 7A] AB PhCO PHCO (corrected) -2.71366 10466755.2 -4.815 -4.780 -2.72912 9969258.9 -4.673 -4.480 -2.80095 7456032.3 -4.496 -4.180 - 2,87614 5441200.7 -4.172 -3,880 -2.98508 2191673.0 -4.016 -3.534 -3.03827 803903.8 -3.238 -3.225 -3.11561 -1408193.9 -2.688 -2.949 -3.15601 -1883755 , 7 -2,464 -2,703 -3,17825 -2339447,3 -2,697 -2,483 -3,19324 -2180978,0 -2,146 -2,287 -3,23511 -2817219,1 -1,776 -2,110 -3,25131 -1754962,1 -1,809 -1,952 -3,30726 -2186721,5 -2,134 -1,808 -3,35919 -1976897,9 -1,429 -1,676 -3,41139 -1204228,2 -1,226 -1,555 -3,49709 -2023429,8 -1,417 -1.442 -3.59266 -1820384.4 -1.532 -1.336 -3.69286 -2324054.5 -1.071 -1.234 -3,80597 -2131118.4 -1.073 -1.136 -3.97078 -2014626.0 -1.119 -1.040 -4.1509 -1795387.1 -1.515 -0.945 -4.47188 -1814828.6 -0.673 -0.850 -4.96056 -1255988.6 -1.020 -0.756 [Table 7B] [Table 7B] AB PhCO PHCO (corrected ) -5.78791 -548412.1 -0.825 -0.660 -7.07361 851318.0 -0.820 -0.563 -9.111182 783459.5 -0,455 - 0.465 -12.14651 4802717.1 -1.636 -0.366 -25.18997 10916603.8 -2.226 -0.267 18.02675 -164173140.9 -99.492 -0.1167 19.58743 4407625.5 1.292 -0.067 10.97926 8262259.0 0.535 0.030 7.07471 6862847.9 -0.053 0. 126 5.06505 6735688.7 0.411 0.217 3.88221 4725572.2 0.533 0.304 3.21722 4883904.6 0, 936 0.383 2.78329 4453119.1 0.509 0.398 2, 43906 4502630.6 0.338 0.413 2.20738 4323592.2 0.293 0.428 2.05995 4636996.2 0.599 0.443 1.95225 4327988.0 0.363 0.458 1.88431 4113055.3 0.468 0.473 1.82583 3925127.9 0.1198.488 1, 79565 4090824.7 0.540 0.503 1.73714 3602165.8 0.532 0.518 1.78714 3602165.8 0.532 0.533 169 In the same manner as in Example 2, the (corrected) o value of PhCO was produced being corrected from all general changes of the PhCO so that the value of PhCO can be continuously changed. A frequency spectrum adjusted for phase 0 order was obtained using the PhCO (corrected) value. The 191-point FFT smoothing filter was applied to the frequency spectrum obtained in order to remove the noise. The signal strength at 100ppm of the frequency spectrum was plotted against each dwell time (τ), thus obtaining longitudinal relaxation curves.

Likewise, in relation to the glycine samples from sample 2 to sample 8, the longitudinal relaxation curves were produced. These results are shown in Table 8a and Table 8B.

[Table 8A] τ Ιητ R c B60 AB RM CM s 0.00002 10.820 -223452548 -113973815 -153789978 -178228228 -104972358 -246578959 -164546450 -110855890 0.00005 -9.903 -220764866 -113476686 -151189303 -175615407 -103741145 242355021 - 161421243, 109523231 0.0001 -9.210 -213785822 -109132248 -145494391 -169383364 -99701346 -232776894 -152766279 -109717316 0.0002 -8.517 -206448001 -104967966 -139986667 -162604599 -95728570 -221446013 -142620729 -99474672 0.0005 -7.601 -197931402 -100547130 -132718041 -154508428 -90629466 -208354032 -130372616 -93503897 0.001 oo -192947884 -97771606 -127810845 -149671180 -87404230 -200170698 -123713038 -89628243 0.002 -6.215 -188920979 -96138064 -125337122 -146261195 -84572235 -192681354 - 119597919 -86955690 0.005 -5.298 -182651499 -95328310 -120529317 -140127445 -81684661 -185584661 -118074661 -83985207 0.01 -4.605 -174673464 -95211875 -114745977 -133418665 -78099599 -176152360 -117899243 -80334886 0.02 -3.912 -158131955 -94813293 -103186001 -119990862 -70179900 -158496672 -116760044 -73208619 0.0333 -3,402 -136479277 -93473268 -89190096 -103695110 -60599394 -135735570 -115512624 -64177531 0.05 -2.996 -111986378 -92169616 -72896931 -83322493 -48943629 -109199937 -113990850 -53830442 0.07 -2.659 -83616271 -91498890 -54326805 -61888457 -35699655 -78834633 -112100357 -42725797 0.1 -2303 -45441923 -89472959 -29221320 -31021625 -18620314 -38067573 -109369439 -2632257 0.13 2.040 -11826080 -88230163 -6321577 -4047329 -2567323 -1216394 106367938 -1116284 0.16 -1.833 18851541 -86408476 14623251 20256674 11387047 32533305 -103347314 891708 0.2 -1,609 55730441 -84055360 38965151 49371738 28609791 71283995 -100042017 15886394 0.24 -1.427 87562319 -82104749 60301841 73891367 43737832 104882886 -96678344 29230004 0.28 -1.273 116104546 -79528254 79044717 96169221 56232291 132647802 -93979203 40473544 0.34 -1.079 152110893 -75783195 103031880 124906019 72743689 171749503 -88650325 55464969 0.4 -0.916 179517812 -72785472 121812946 148166 521 85797098 201118735 -84170877 66958048 0.5 -0.693 217116419 -67506536 146125004 177185971 102290755 239283701 -76427493 83488895 0.63 -0.462 249137168 -60897291 167678017 202539042 116657881 273352679 -66770571 96802344 0.8 -0.223 275501046 -52017966 184191946 222538656 128325242 299508721 -55681124 108087289 53 τ In τ R c B60 AB RM CM s 1 0,000 291829682 -42618969 195925467 235659237 135532234 316575919 -42470387 116083938 1.2 0.182 300256959 -32698550 201548659 242688632 139932258 325185911 -31615145 121447452 1.4 0.336 304940602 -24209943 204820673 245893913 142242613 329995836 -20061628 123886477 1.7 0.531 307848790 -11918668 206835033 248369851 143520726 333431390 -3673180 127135288 2 0.693 308983597 161159 206421658 249206044 144491414 335334199 9870489 128909196 2.4 0.875 310445549 15231321 208795230 249848942 145496375 336432566 28112821 131501519 2.8 1.030 310947396 27793201 209159901 249979498 145379413 335330533 45396951 133530821 3.33 1,203 3106 210714058 54927 43188908 250495564 146006546 336681032 65164019 135229935 4 1.386 311571293 60898774 210677854 250809811 146335767 336665750 86187941 137458635 4.8 1.569 311781927 78226211 211439853 250627829 146800767 337418406 107885087 139164470 5.1 1.740 312366530 94478646 211327024 250865153 146944574 337460405 127837716 141752403 8.7 1.902 312518491 109788512 212477081 251059662 147345928 339521961 145959475 143414800 8 2.079 312882730 125459886 212165002 250913879 147769648 340115510 163015193 145066258 9.5 2.251 312919959 138647038 212773247 250965416 148624774 339047018 177998842 146464492 11.3 2.425 313148134 148583610 213621765 250972876 148637428 339312823 189233962 147952404 13.5 2.603 313050094 157178025 213944848 250824881 148972925 340599719 198361336 148970392 16 2,773 312696454 162353123 213768759 251096508 149040386 339743597 204352768 149645023 20 2,996 313460833 167277368 214162944 250875496 148988150 339599694 208849756 150485508 30 3.401 313132829 1700 76590 214191805 251668512 148492521 340080668 211081068 151491570 45 3,807 313588462 170437341 214210897 251911385 149339653 341044614 211579762 152491121 90 4,500 314249774 170971176 214106925 251423359 149123683 339633998 211407963 153735886 54

The longitudinal relaxation curves obtained as Table 8A and Table 8B for the eight glycine samples were analyzed using the following formula.

[Formula 5] J / vd- ^ Vc = l

Four types of glycine constituents were found (in the formula, n = 4). Table 9 shows the results of calculating the estimated value and standard error of the longitudinal relaxation time Si.

[Table 9]

Estimated value Standard error SI 1.47435 0.00294 S2 -1.23516 0.00145 S3 -8.62947 0.07122 S4 -6.76086 0.13573 Table 10 shows the results of calculating the estimated value of the coefficient fi e of the estimated value of the constant C.

[Table 10] (Estimated value) RC B60 AB RM CM CM 3922485.6 132436248.0 4845862.1 1966110.7 3421986.3 3534487.9 160862124.9 17783980.6 f2 248580234.3 0.0 165025866.7 197161612 , 9 113595157.9 262613325,4 5485826,2 101211592,1 f3 13665971,4 7488055,0 10813062,3 12107575,6 7197262,2 18532940,9 19078867,5 8871063,2 f4 4251969,2 2482622,1 4609970,4 5192533 , 7 3783701.5 10983654.8 6317097.2 4522775.7 C 43513387.8 26246043.0 28971277.6 35191516.3 21060048.4 44430862.0 21673356.3 19006253.5 Table 11 shows the results of the standard error calculation of the coefficient fi and the standard error of the constant C.

[Table 11] (standard error) RC B60 AB RM CM CM 272398.7 261636.2 259644.0 263914.2 254269.2 274354.3 253067.3 253308.7 f2 268333.2 137389.4 267470.3 267977 , 6 266844.8 273505.6 306494.6 267576.6 f3 867923.3 760065.1 867058.6 904499.8 809440.1 1322765.3 1025698.5 851605.7 f4 812400.7 612089.3 774466.6 824241 , 1 681169.9 1259579.6 1029641.4 737513.3 C 390317.9 404307.8 378943.8 382350.8 370697.5 399695.0 418530.0 373790.3 (Λ

Ui 56 From these results, the TI longitudinal relaxation time values of the glycine constituents were determined as Tl = 4.37s (assigned to type γ), Tl = 0.29s (assigned to type a), Tl = 1, 15ms (unknown assignment), and Tl = 0.18ms (unknown assignment).

In addition, the ratio of the constituents of each sample was determined as in Table 12, using the coefficient of force fi. Additionally, the confidence interval of ± 95% was calculated at the same time.

[Table 12] ΤΙ RC B60 AB RM CM S 4,37s 1,4510,20¾ 93,0010,36¾ 2,6210,28¾ 0,9110,24¾ 2,6710,39¾ 1,2010,16¾ 83,8910,26¾ 13 , 43 + 0.38¾ 0.29s 91.9210.20¾ 0.0010.19¾ 89.0610.28¾ 91.1010,24¾ 88.75v0.41¾ 88.8210,18¾ 2.8610.31¾ 76.45 + 0, 40¾ 1.15ms 1.5710.59¾ 1.7410.85¾ 2.4910.82¾ 2.4010.75¾ 2.9611.05¾ 3.7110.84¾ 3.29 + 1.06¾ 3.42 + 1.10¾0, 18ms 5.0510.63¾ 5.22611,05¾ 5.8410.92¾ 5.5910.82¾ 5.6211,24¾ 6.2710.88¾ 9.95 + 1.05¾ 6.70 + 1.27¾ (Estimated Range Value of 195 de confidence) ΙΛ 58

In this way, it was possible to determine the ratio of the constituents of the glycine solid crystal polymorphisms with high precision. The longitudinal relaxation time of each constituent, as well as the ratio of the constituents, could be measured at the same time.

Lisbon, July 13, 2010

Claims (4)

A method of measuring proton nuclear magnetic resonance in a solid sample placed in a sample tube (21) made of glass, the inlet of which is closed with a cap (23), the method comprising the steps of: applying a pulse of inversion to said solid sample; applying a reading RF pulse to said solid sample after the inversion recovery delay time lapse; receive a FID signal after the lapse of the delay time of the reception Dd after the end of the pulse RF reading application; obtaining a proton magnetic resonance spectrum by subjecting the FID signal to a Fourier conversion process; dehydrating the solid sample during measurement by connecting a small hole made in the cap 23 of the sample tube 21 with a tube 24 leading to a desiccant held in a vessel. A method of measuring nuclear magnetic resonance for solid samples according to claim 1, wherein the inversion pulse is a 180 degree pulse, and the RF reading pulse is a 90 degree pulse. A nuclear magnetic resonance measurement method for solid samples according to claim 1 wherein the reception delay time Dd is set to a value ranging from 5 microseconds to 20 microseconds. 2 The solid magnetic resonance measurement method for solid samples according to claim 3, wherein the receive delay time Dd is set to a value ranging from 10 microseconds to 15 microseconds. Lisbon, July 13, 2010 1/6 23 2/6 23 2/6 FIG. 2 FIG. 3 3/6 REPEAT N 90 " 'Dei < μ RD TRANSMITING! Fia 4 (b) FAST Y) Î ± 1 and β1 β2 0.0 0.4 0.4 GB 0.8 1.0 1.2 4/6 5/6 you HE The time (sec) is 0 ° C. 7 Mz '